ReCoG: Relational and Compact Context Graph Learning for Few-shot Molecular Property Prediction

TL;DR

ReCoG introduces a novel framework for few-shot molecular property prediction that effectively models relational context and suppresses irrelevant information to improve molecule representations.

Contribution

It proposes a joint relational and compact context graph learning framework with two core modules, enhancing context exploration and information utilization in FSMPP.

Findings

Theoretical demonstration of joint relational and compact knowledge extraction importance.

Effective suppression of irrelevant signals improves molecule representation.

Enhanced performance in few-shot molecular property prediction tasks.

Abstract

Few-shot molecular property prediction (FSMPP) is essential in drug discovery and materials design, where high-quality labeled data are often scarce and expensive to obtain. Despite the promising performance of existing methods, especially context-aware methods, they still face two-fold severe challenges with \textit{insufficient structural context modeling} \& \textit{redundant auxiliary context learning}, leading to inadequate context graph exploration and ineffective information utilization for effective molecule representation learning. To address these, in this paper, we propose a novel framework by learning on \textbf{\underline{Re}}lational and \textbf{\underline{C}}ompact c\textbf{\underline{o}}ntext \textbf{\underline{G}}raph, named \textbf{\method}, to comprehensively exploit the context graph for expressive molecular property prediction. Specifically, the proposed \method contains two core modules: a \textbf{(1) cross-property relational learning module} to better model the structural and relational context information, and a \textbf{(2) context graph information bottleneck module} to adaptively suppress irrelevant auxiliary signals for compact context information utilization, followed by a detailed theoretical demonstration regarding the importance of joint relational and compact knowledge extraction in context graphs.